Full text - Acta Palaeontologica Polonica

Patterns of larval development in Cretaceous pipid frogs ZBYNĚK ROČEK and EDUARD VAN DIJK Roček, Z. and Van Dijk, E. 2006. Patterns of larval development in Cretaceous pipid frogs. Acta Palaeontologica Polonica 51 (1): 111–126. A developmental series of nearly 250 tadpoles of Shomronella jordanica from the Early Cretaceous (Hauterivian) of the Shomron (Samaria) region of central Israel, a small collection (12) of Thoraciliacus rostriceps tadpoles from the Early Cretaceous (Aptian) of Makhtesh Ramon, Israel, and 13 tadpoles from the Late Cretaceous (Campanian/Maastrichtian boundary) of Stompoor, Marydale, South Africa were studied. These were compared with published data on the develop− ment of Paleogene (Middle Eocene–Early Oligocene) pipids from Patagonia, and with normal development of the con− temporary pipid frog Xenopus. The comparisons of the developmental series of the Cretaceous and contemporary pipid frogs provided data on changes in the developmental pattern, namely of the degree of ossification and other developmen− tal events. In general, it seems that ossification has become considerably delayed in the course of pipid evolution, whereas most anatomical features typical for free living pipid larvae were well established as early as in the Early Cretaceous. Comparisons with the developmental series of specimens from the Late Oligocene Palaeobatrachidae (closely related to the Pipidae) from Bechlejovice near Děčín, Czech Republic revealed that some morphological differences between the two families might be explained by their developmental mode (e.g., formation of the opisthocoelous vertebral centrum in Shomronella). The uncinate process on three anterior pairs of ribs in Shomronella is the character retained in primitive anurans, and indicates close phylogenetic relations of early pipoids to discoglossoids. Key words: Anura, Pipidae, larval development, Cretaceous, Oligocene, Israel, South Africa. Zbyněk Roček [[email protected]] Laboratory of Palaeobiology, Geological Institute, Academy of Sciences, Rozvojová 135, CZ−165 00 Prague 6, Czech Republic and Department of Zoology, Charles University, CZ−128 44 Prague 2, Czech Republic; Eduard Van Dijk [[email protected]] Department of Plant and Animal Biology, University of Stellenbosch, Private Bag X1, Matieland 7602, South Africa.

Introduction Frogs related to extant Pipidae are known from as early as the Early Cretaceous of Israel (Neocomian, at least 122 Ma). They are the earliest anurans known in which tadpoles were pre− served. A series of tadpoles (nearly 50), not associated with adults except for a fragmentary hind leg, HUJZ 12029, comes from the Lower Cretaceous (Hauterivian; Lang and Mimran 1985) of Shomron (Samaria) region in central Israel. These were described as Shomronella jordanica Estes, Špinar, and Nevo, 1978, and their ossification sequence and growth rates were later investigated, together with a large additional series of nearly 200 specimens collected in 1990s, by Chipman and Chernov (2002). Unfortunately, these specimens represented mostly a single metamorphic stage (with few early larvae and one young postmetamorphic individual). Despite unequal rep− resentation of developmental stages in the sample, because of the preservation of soft parts of body, one can recontruct the main anatomical transformations in the course of Shomro− nella’s development, and not only the ossification sequence. Besides Shomronella Estes, Špinar, and Nevo, 1978, less numerous but a representative developmental series was rec− ognized in the large collection of adult frogs from the Early Cretaceous of Makhtesh Ramon, Israel. This collection was first dated as Neocomian by Nevo (1968), Barremian by Estes Acta Palaeontol. Pol. 51 (1): 111–126, 2006

et al. (1978), and Aptian by Gvirtzman et al. (1996). Tadpoles from Makhtesh Ramon were noted, but not described, by Nevo (1956). Because these tadpoles represent the earliest stages of ossification as well as late metamorphic and early postmetamorphic stages, they can be easily associated with adults of Thoraciliacus rostriceps Nevo, 1968 (mostly on the basis of the distinctive shape of the parasphenoid and sacral and pelvic features). Moreover, specimen HUJZ−Th04 bears an imprint of a tadpole on the upper surface of a slab (Fig. 1E) and a fragmentary posterior limb of an adult on the lower side, which helps associate the tadpoles with adults. During the past several decades, large series of pipid frogs, including tadpoles, were recovered from the marginal facies deposits of a crater lake dated to the Campanian/Maastrichtian boundary around 74 Ma (Smith et al. 2002) on the Stompoor farm, Marydale, South Africa. A preliminary report was pub− lished by Van Dijk (1985). Excavations made by Roger Smith in 1999 yielded 243 specimens that were deposited in the South African Museum, in Cape Town. Earlier findings (Geo− ffrey Grantham in 1970s and Roger Smith in 1984) comprised several tens of specimens that are now deposited in the collec− tions of the Department of Plant and Animal Biology, Univer− sity of Stellenbosh and in the South African Museum (SAM material is from borehole Stompoor 109). In the early 1980s part of this older material was sent to Richard Estes but the http://app.pan.pl/acta51/app51−111.pdf

112

ACTA PALAEONTOLOGICA POLONICA 51 (1), 2006

Fig. 1. Larval development in the Early Cretaceous pipid Thoraciliacus rostriceps from Makhtesh Ramon, Israel. A. The earliest recorded larva, with rudimen− tary ribs and not−yet fused neural arches; stage NF 59, dorsal aspect (HUJZ−Th01). B. Same developmental stage (HUJZ−FL3b). C. Stage NF 60, ventral view (HUJZ−Th03); note the anterior tip of the parasphenoid exceeding beyond the both frontoparietals. The photo taken by Zeiss Stemi 2000C stereomicroscope (C1). The photo taken by means of the image analysing software “Image Pro Plus“ (C2). D. Stage NF 60, dorsal view (HUJZ−Th02). The photo taken by Zeiss Stemi 2000C stereomicroscope (D1). The photo taken by means of the image analysing software “Image Pro Plus“ (D2). E. Stage NF 61, ventral view (HUJZ−Th04). Rudiments of ilia and femora are marked by arrows. F. Estimated stage NF 63, according to ossified tips of toes, probably ventral aspect (HUJZ−Th11). G. Postmetamorphic stage with the ilio−sacral articulation, ventral aspect (HUJZ−F301). H. Postmetamorphic stage, ventral aspect (HUJZ− F235); note fusion of sacral and praesacral diapophyses. I. Adult, ventral aspect (HUJZ−F93; holotype, cf. also Trueb 1999). Scale bars 5 mm.

study was not completed because of his death, and this mate− rial is apparently lost. Frogs from Stompoor were recently de− scribed by Trueb et al. (2005). They are generally not well pre− served, and the bones are usually shattered, apparently by crystallization of minerals. In spite of this, the series includes some adults in which important anatomical details may be ob− served, as well as tadpoles and metamorphosing froglets, which provide data on premetamorphic development. The principal aim of the present study is to compare de− velopmental patterns among these three Mesozoic pipid taxa and compare them with the development of later representa− tives of the pipid lineage in order to resolve evolutionary trends in pipid ontogeny. The ontogeny of Cenozoic pipids was documented from a series of 12 tadpoles of Llanki− batrachus truebae Báez and Púgener, 2003 (Middle Eocene– Early Oligocene of Pampa de Jones and Confluencia sites, northwestern Patagonia, Argentina; Báez and Púgener 2003). This, together with extensive knowledge of the development

in contemporary Xenopus (Nieuwkoop and Faber 1967; Trueb and Hanken 1992) makes it possible to reconstruct principal changes in pipid development in the course of more than 120 million years of their history and to assess the evolutionary trends of their ontogenies. Also, comparisons of Mesozoic pipids with a large devel− opmental series of palaeobatrachids from the Late Oligocene of the Czech Republic (Roček 2003b) provides information that can help to solve the problem of pipid−palaeobatrachid relationships.

Material and methods Developmental series of Thoraciliacus is represented by 12 larval and metamorphic specimens (HUJZ Th01–Th11, FL3a, 3b), and several specimens (both catalogued and un−

ROČEK AND VAN DIJK—LARVAL DEVELOPMENT IN CRETACEOUS PIPID FROGS

113

Fig. 2. Larval development of the Early Cretaceous pipid Shomronella jordanica from the Shomron region, Israel. A. The earliest recorded larva, stage ap− proximately NF 47–50, probably dorsal aspect (HUJZ−13150). Displaced eyeball marked by arrow. Note complete parasphenoid. The photo taken by Zeiss Stemi 2000C stereomicroscope (A1). The photo taken by means of the image analysing software “Image Pro Plus“ (A2). B. Moderately older stage, approxi− mately NF 51–54, dorsal view (HUJZ−13062). Displaced eyeball marked by arrow. C. Approximately same stage as the previous, dorsal view (HUJZ−13190). Rudimentary frontoparietals. D. Approximately same stage as the previous, dorsal view (HUJZ−13132). Scale bars 5 mm.

catalogued) illustrate postmetamorphic development. Shom− ronella is represented by a series of 44 tadpoles, most involv− ing their counterparts (HUJZ 12001−12044); this series was collected in mid−1970s (Estes et al. 1978), and includes the holotype (HUJZ 12020). The second series was collected be− tween 1989 and 2000 (Chipman and Tchernov 2002), and is represented by nearly 200 tadpoles (HUJZ 13001−13197). Thoraciliacus and Shomronella are deposited in the collec− tions of the Department of Evolution, Systematics and Ecol− ogy, Hebrew University, Jerusalem. The pipid from Stompoor was represented by 13 tadpoles (SAM−K6466, K9556, K9559, K9560, K9583, K9586, K9594, K9604, K9656, K9662, K9673, and K9679—2 indi− viduals on a single slab), deposited in the collections of the South African Museum in Cape Town and 2 tadpoles (USP 484 and 485) deposited in the collections of the Department of Plant and Animal Biology, University of Stellenbosch. Comparisons with Xenopus laevis (Daudin, 1802) were based on a series of several tens of individuals prepared as cleared−and−stained whole mounts covering the period be− tween stages 46 and 66 of Nieuwkoop and Faber (1967). This material is deposited in the Department of Zoology, Faculty of Natural Sciences, Charles University, Prague. Comparisons with Palaeobatrachus were based on a se− ries of 171 individuals representing larval, metamorphic and postmetamorphic stages recovered from the Late Oligo− cene of Bechlejovice, Czech Republic (see details in Roček 2003b). Where possible, identification of stages was made on the basis of external characters (such as development of limbs) in order to avoid individual variation of internal criteria. Some individuals from the Jerusalem collections were inves− tigated and documented using “Image Pro Plus“ software which made it possible to emphasize contrasts between the skeletal structures and surrounding matrix.

Institutional abbreviations.—DZ−FNSP, Department of Zool− ogy, Faculty of Natural Sciences, Charles University, Prague, Czech Republic; HUJZ, The Department of Evolution, Sys− tematics and Ecology, Hebrew University, Jerusalem, Israel; SAM, South African Museum, Cape Town, South Africa; USP, Department of Plant and Animal Biology, University of Stellenbosch, South Africa. Other abbreviation.—NF, stages of normal development of contemporary pipid frog Xenopus laevis, as defined by Nie− uwkoop and Faber (1967).

Results Development in the Early Cretaceous pipid Thoraciliacus sp. (Fig. 1).—Thoraciliacus is the only Early Cretaceous pipid in which the adult may be associated with larval and metamor− phic stages, principally on the basis of two facts: (1) the shape of the parasphenoid is the same in adult Thoraciliacus (Nevo 1968: fig. 10C) and in all tadpoles from the same locality in which this bone is preserved, and (2) specimen HUJZ−Th04, which includes both the skeleton of a tadpole as well as part of the adult skeleton (hind extremity fitting into the variation range of Thoraciliacus; Nevo 1968) on a single slab. Although there are only 12 larvae preserved, they repre− sent at least three developmental stages and thus provide im− portant ontogenic information. The youngest preserved stage (HUJZ−Th01) is a complete skeleton in dorsal aspect (Fig. 1A), including the frontoparietal, which is still divided by a fontanelle anteriorly. Its parietal sections are fused with each other posteriorly, possibly leaving an indistinct median su− ture. The parasphenoid was either not yet developed or (more probably) it was not preserved. The medial walls of the otic capsules had only begun to ossify. There are nine vertebrae http://app.pan.pl/acta51/app51−111.pdf

114

developed and at least five posterior of them with separate neural arches. Three pairs of ribs (on the 2nd–4th vertebrae) are present. The limbs are not yet ossified. Considering only the degree of ossification, this larva is at stage NF 54. However, comparison with the normal development of Xenopus laevis (Nieuwkoop and Faber 1967) suggests that limbs were probably developed but with only a cartilaginous skeleton (Fig. 4C); thus from the inferred external appear− ance this stage could correspond to stage NF 59. HUJZ−FL3a and b (part and counterpart) represent about the same devel− opmental stage as judged by nine vertebrae and three pairs of ribs; and its postsacral vertebrae were not yet ossified. The next stage is represented by specimens HUJZ−Th02, Th03, and Th05. HUJZ−Th02 is in the dorsal aspect, the other two are in the ventral aspect. The parasphenoid is well devel− oped, anteriorly exceeding the plane of the frontoparietals, which are fused to various degrees. The otic capsules were well ossified (HUJZ−Th03) or at least calcified. Ten verte− brae may be discerned with their neural arches still separated from one another. The clavicles are present (HUJZ−Th02, Th03). Although limbs are not present, it is possible that this may be due to preservation status. This seems to be evi− denced by specimen HUJZ−Th04 (Fig. 1E) in which the ilia are separated from one another and the femora are preserved, although other features (especially number of vertebrae) are the same as in other specimens of this stage. A typical feature is that the vertebral column is shorter than the skull measured to the rostral tip of the parasphenoid. These larvae may be roughly identified as stage NF 60–61. Specimen HUJZ−Th11 (Fig. 1F) represents a stage in metamorphosis (probably 63, judging by degree of ossifica− tion of the ilia in Xenopus), with well developed hind limbs, but the ilia are not yet articulated to the vertebral column. The sacral vertebra does not have diapophyses. Postmetamorphic development is characterized by addi− tional pair of ribs on the 5th vertebra (but not on the atlas) and occasional fusion of diapophyses of the the sacral and the most posterior presacral vertebrae (Fig. 1H). Development in the Early Cretaceous pipid Shomronella jordanica Estes, Špinar, and Nevo, 1978 (Figs. 2, 3, 5).— The earliest larva in the developmental series is HUJZ−13150 (Fig. 2A) in which only the parasphenoid is ossified. It should


be noted that Chipman and Tchernov (2002) illustrated a still earlier tadpole, about 4 mm long with no ossification (see their fig. 2), but because they did not refer to the fossils by cata− logue numbers, we were not able to relocate this specimen. Based on the external body shape, this individual may be in stages NF 47–50. The tail is not complete but the length of the trunk to the plane of the vent is about 10 mm. The anterior part of the parasphenoid is clearly delimited, which means that it was well ossified; its anterior tip reached beyond the plane of the eyes. The frontoparietals begin to ossify as faint strips adjoin− ing the constriction of the posterior half of the parasphenoid and soon extend anteriorly. Simultaneously, two ossification centers lateral to the posterior end of the parasphenoid (termed the exoccipitals by Chipman and Chernov 2003) and 6–8 pairs of neural arches can be seen. Externally, these lar− vae (HUJZ−13062, 13190, 13132; Fig. 2B–D) do not differ much from the previous stage, except for size, but the prog− ress in ossification suggests a more advanced stage (perhaps NF 51–54 based on distinct outlines of the body without pos− terior limb buds; Fig. 2B). Curiously enough, these larvae are characterized by pigmented eyes extruded from the outlines of the head (marked by arrows in Fig. 2A1, B), which sug− gests that the head skeleton was not yet well chondrified. Also, the head in these early larvae is not oval (though more or less symmetrical) in shape, and their tail was apparently (as suggested by a slightly older larva HUJZ−12036; Fig. 3A1) shorter than the body. The next stage (about NF 55) is represented by HUJZ− 12036 (Fig. 3A). Compared with the previous one, it had three pairs of ribs, although it seems that the number of verte− bral rudiments did not increase. Faint imprints of cartilages suggest that the larval skull was already well chondrified; the lower jaw consisted of the infrarostral (possibly still paired) and Meckel’s cartilages, both articulated with each other (compare Fig. 3A3 and Fig. 3O). Interpretation of blood ves− sels and nerves by Jarošová (in Estes et al. 1978: fig. 2A) is largely confirmed, possibly except for what she identified as the vena capitis lateralis; our examination of the specimen suggests that it might be the outer outlines of the cartilagi− nous otic capsules. Further, the optic nerve extending to the inner surface of the eye (only slightly protruding from the

Fig. 3. Larval development of the Early Cretaceous pipid Shomronella jordanica from Shomron region, Israel (continued). A. Stage NF 55, dorsal view (HUJZ−12036). The photo of complete specimen taken by Zeiss Stemi 2000C stereomicroscope (A1). Imprint of parasphenoid taken by Zeiss Stemi 2000C stereomicroscope (A2). The photo of the head with vessels and nerves taken by means of the image analysing software “Image Pro Plus“ (A3). B. Stage NF 59−60, head and anterior part of vertebral column, dorsal aspect (HUJZ−13056). Note jaw cartilages comparable with those in corresponding stage of Xenopus (O); otic capsules and vertebrae are covered by thin layer of sediment. C. Stage NF 60 in dorsal view (HUJZ−12012). Angulosplenial on the left side marked by arrow. Otic capsules still incompletely ossified. D. Stage NF 60, head in dorsal view; nasals marked by arrow (HUJZ−12019). E. Stage NF 60, dorsal wiev (HUJZ−12013). V1 and V2 fused. Rib with uncinate process marked by arrow. F. Stage NF 60, dorsal view (HUJZ−13184). Hind limbs rudimentary and widely separated from vertebral column. G. Stage NF 61, dorsal view (HUJZ−12009). H. Stage NF 63, ventral view (HUJZ−12021). Note well developed but displaced hypochord. I. Stage NF 63, ventral aspect (HUJZ−12017a). Note development of the vertebral centra (ossification within notochordal canal marked by arrow). The photo taken by Zeiss Stemi 2000C stereomicroscope (I1). The vertebral column taken by means of the image analysing software “Image Pro Plus“ (I2). J. Stage NF 63, dorsal view (HUJZ−12003). The neural arches of the posterior vertebrae are broken off so the inorganic filling of the vertebral canal is visible. V1 and V2 still separate. The photo of the complete specimen taken by Zeiss Stemi 2000C stereomicroscope (J1). The vertebral column taken by means of the image analysing software “Image Pro Plus“ (J2). K. Stage NF 63–64 in ventral view (HUJZ−13136). L. Stage NF 63–64 in dorsal view (HUJZ−13073). M. Stage NF 63−64 in ventral view (HUJZ−13080). N. Early postmetamorphic individual, probably in ventral view (HUJZ−13053). Iliosacral articulation established. O. Xenopus laevis, stage NF 57, skull in dorsal view (DZ FNSP XL57−ex2). Scale bars 5 mm.

®


115

http://app.pan.pl/acta51/app51−111.pdf

116

head outlines) indicate that the pigmented area represents the entire eye, not only the lens (compare with Fig. 3O where the eye lens is separated from the cartilaginous eye cup). The subsequent stage is represented by specimen HUJZ− 13056 (Fig. 3B). It differs from the previous stage by having nine vertebrae and ossified posteromedial parts of the otic capsules. Meckel’s cartilage and the internasal plate are com− parable with those in stage NF 59 of Xenopus (Fig. 3O), and the lower jaw still lacks dermal ossification. The first dermal ossifications after the parasphenoid and frontoparietals are the angulosplenials adjoining Meckel’s cartilages on their medial side, which appear only in stage NF 60 (HUJZ− 12012, Fig. 3C; HUJZ−13040); the frontoparietals are still separated along their entire length. Nine to ten vertebrae may be recognized, the neural arches on both sides of anterior ver− tebrae are completely fused, while the more caudal trunk ver− tebrae are still separated in the midline. The first and second vertebrae are already fused into a single unit; fusion is docu− mented by HUJZ−12013 (Fig. 3E) where the first and second vertebrae still may be recognized on the right side, and by HUJZ−1203 (Fig. 3J1, J2). There are three pairs of ribs articu− lated with the original vertebrae 2–4. It is noteworthy that the second pair of ribs bear a distinct uncinate process on its pos− terior side, as documented by HUJZ−12013 (Fig. 3E) and nu− merous other specimens of this and an older age (HUJZ− 12007, 12010, 12017a, 12021, 12026, 12038, 13005, 13043, 13048, 13063, 13170, 13184). Still in stage NF 60, the tiny rudiments of the nasals may be recognized (HUJZ−12019, Fig. 3D). Appearance of these dermal ossifications is accom− panied with an increase in size of the frontoparietals so that the two bones approach one another and may ultimately fuse in their posterior parts (HUJZ−12037). The earliest ossified rudiments of the posterior limbs (with no pelvic elements) also appear in stage NF 60 (HUJZ− 13184, Fig. 3F). The diaphyses of the tibia and fibula (well separated) may be recognized and the elements of the pecto− ral girdle become ossified (among the first is the scapula). On the other hand, the hypochord is still absent. Stage NF 61 (HUJZ−12009, Fig. 3G) is characterized by the appearance of the hypochord and rudimentary ilia. The hypochord usu− ally appears simultaneously with the first caudal (i.e., 10th) vertebra. The great majority of the specimens are in stage NF 63. These tadpoles still have a long tail (Fig. 3H), the hind limbs


are ossified up to the distal tarsal elements, and the halves of the pelvis are in contact with each other. The forelimbs are ossified up to the distal ends of the separate ulna and radius. In specimen HUJZ−12021 the columella is clearly visible. In all individuals of this developmental age, in which the verte− bral column is exposed from the ventral side, one can see that the centra were ossified at the bases of the transverse pro− cesses and on the dorsal part of the notochordal canal (epi− chordal ossification). This condition results in a longitudinal groove on the ventral side of the vertebral column (Fig. 3H, I1, M). The groove in the anterior vertebral centra was soon filled with ovoid calcifications or ossifications that replaced the notochord. Image analysing software reveals that those parts of the fillings corresponding to each centrum are sepa− rated from the lens−shaped fillings in the intervertebral spaces. The latter fillings fuse later with those within the ad− jacent posterior centrum, thus giving origin to the opistho− coelous centrum (Fig. 3I2; see also Estes et al. 1978: fig. 6C). These fillings replacing the notochord should not be con− founded with the inorganic fillings of the vertebral canal; these can be observed only from the dorsal side when the neural arches are broken off (Fig. 3J 1, J2). Further development may be characterized by gradual elongation of the ilia and their placement into an ultimate po− sition nearly parallel with the axis (compare Fig. 3L with Fig. 3M, N). It is obvious that the hypochord was separated from the caudal vertebrae in the late stages of metamorphosis and that the tail was still as long as the hind limbs (Fig. 3M). Development in a South African Late Cretaceous pipid (Fig. 4).—The earliest larva (specimen USP 484; Fig. 4A), based on the presence of paired frontoparietal, is in stages NF 54 or 55; since Nieuwkoop and Faber (1967) characterized stage 56 by fused frontoparietals. However, they based their stages of this developmental period on external characters, namely on the relative size of the extremities. In our material of Xenopus, dermal ossification was considerably retarded and separate frontoparietals were still observed in stage 59 which is characterized by the extended forelimbs reaching to the base of the hind limbs. We believe that the body shape (including size of limbs) is more important for determining developmental stages than the degree of ossification, which may vary because of heterochrony (see Roček 2003b for dis− cussion of variation of the internal characters in Xenopus), so we determined this larva as stage NF 59.

Fig. 4. Larval development in the Late Cretaceous pipid from Stompoor, South Africa, compared with normal development of Recent Xenopus laevis. A. The earliest recorded larva, stage NF 59, dorsal aspect (USP 484). Frontoparietals feebly ossified. B. Slightly older individual, stage NF 59–60, ventral aspect (SAM−PK−K9560). C. Xenopus laevis, stage NF 59 (DZ FNSP XL59−ex2). D. Beginning of metamorphosis, stage NF 60, ventral aspect (SAM−PK−K9656). E. Stage NF 60, ventral aspect (SAM−PK−K6466). Frontoparietal well ossified but still paired. F. Estimated stage NF 63, ventral aspect (SAM−PK−K9583a). G. Counterpart of the same specimen, dorsal aspect (SAM−PK−K9583b). Scapula marked by arrow. H. Stage NF 63, ventral aspect; note relative size of the parasphenoid (SAM−PK−K9662a). Scapula marked by arrow. I. Stage NF 63, ventral aspect (SAM−PK−K9662b). Scapula marked by arrow. J. Stage NF 63, dorsal aspect (SAM−PK−K9604a). K. Counterpart of the same specimen, ventral aspect (SAM−PK−K9604b). L. Recent Xenopus laevis, stage NF 63 (DZ FNSP XL63−ex3). M. Recent Xenopus laevis, stage NF 60, skull in ventral view (DZ FNSP XL60−ex2). N. Estimated stage NF 63–64, dorsal aspect (SAM− PK−K9620). Angulosplenial marked by arrow. O. End of the metamorphosis, stage NF 66, ventral aspect (SAM−PK−K9698). P. Adult, dorsal aspect (SAM− PK−K9610). Q. Adult, ventral aspect (SAM−PK−K9666). Arrow marks level of the jaw joint. R. Late metamorphic stage, dorsal view (SEM−PK−K9559a). S. Counterpart of the same specimen, ventral view (SEM−PK−K9559b). T. Disarticulated sacral vertebra with coalesced urostyle (SAM−PK−K9523). U. Adult, part of the skull in ventral view (SAM−PK−K9651). V. Tadpole, supposedly from the Cretaceous of Zaire (USP 493). Scale bars 5 mm.

®


117

http://app.pan.pl/acta51/app51−111.pdf

Randon sibiricus

Salamandrella keyserlingi

Ambystoma mexicanum

Shomronella jordanica Palaeobatrachus sp.

Xenopus laevis

Fig. 5. Comparison of the relative sequence of some developmental events of the skull in the Cretaceous Shomronella, Oligocene Palaeobatrachus, and Recent Xenopus, with the Paleozoic Apateon (Temnospondyli: Branchiosauridae), primitive recent caudates Ranodon and Salamandrella (Cryptobranchoidea: Hynobiidae), and neotenic caudate Ambystoma (Ambystomatidae). Bones that are re− tained in the anurans are in bold. Because of differences in definition of metamorphosis among various authors, and because of different definition of developmental stages in the Paleozoic amphibians, caudates and anurans, comparison in terms of exactly corresponding anatomical stages is not possible. The most objective for staging in the amphibians is formation of the mouth, beginning of metamor− phosis (associated with reduction of gills in caudates, development of the limbs in anurans), and the end of metamorphosis (loss of tail in the anurans). Data on Apateon from Schoch (1998), on the caudates from Lebedkina (2004), and on Palaeobatrachus from Roček (2003b).

Apateon pedestris

118 ACTA PALAEONTOLOGICA POLONICA 51 (1), 2006


This larva is characteristic by, besides paired fronto− parietal, the posterior part of the otic capsules being ossified and by eight pairs of neural arches, all separated at the midline. It possesses two (plus developing 3rd) pairs of ribs on vertebrae 2–4. A slightly more advanced larva (stage NF 59–60, sug− gested by its larger size, number of vertebrae and degree of ossification of the otic capsules), is preserved in ventral as− pect (Fig. 4B); it has eight (plus developing 9th) vertebral centra. Frontoparietals are more massive but still separated by a fontanelle along their entire length, and the otic capsules are ossified in their medial parts. There are three pairs of ribs (the most posterior of them is still rudimentary). Specimens SAM−PK−K9586 and SAM−PK−K9594 are at the same stage. The next stage (Fig. 4D) is characterized by early ossifi− cation of the rudiments of the ilia and femora (not included in Fig. 4D), more ossification of the frontoparietals, which are close to each other but still separated anteriorly, otic capsules ossified laterally and ventrally, and ribs clearly defined. Ten vertebral centra may be discerned, plus two (11th and 12th) rudimentary caudals. Based on the rudiments of the pelvis and hind limbs, it can be determined as stage NF 60 (see Roček 2003b: fig. 1M). This developmental stage is docu− mented also by specimen SAM−PK−K9556. The most frequent stage is about NF 63, based on the de− velopment of the ilia and posterior limbs (Fig. 4E–K). It may be characterized as an advanced metamorphic stage with a tail reduced considerably. The two frontoparietals are either sepa− rated by a midline suture or coalesced together. A remarkable feature of the frontoparietal is that it is depressed along the midline, a character maintained in adults. Its anterior margin is convex, with the tip of the parasphenoid extending from below it (Fig. 4H), as is the case with contemporary Xenopus in stage NF 60 (Fig. 4M). The diaphyses of the long bones of the poste− rior extremities are ossified, as are the distal digits. The ante− rior tips of the ossified portions of the ilia are posterior to the level of the most posterior vertebra (11th or 12th), and the two halves of the pelvis are still moderately separated. Develop− ment of the forelimb is considerably retarded, as inferred from the fact that only the scapula is ossified. A rudimentary hu− merus is present in SAM−PK−K9604 (Fig. 4J, K). In spite of the advanced development of the postcranial skeleton, the der− mal ossification remains confined only to the frontoparietal and parasphenoid. Only in a slightly later stage (NF 63–64; Fig. 4N) can one observe the earliest rudiments of the dermal ossification of the lower jaw (angulosplenial). The two halves of the pelvis are in contact with one another by means of ossi− fied ischia. This developmental stage is documented also by SAM−PK−K9673a, b and SAM−PK−K9679. At the end of metamorphosis (stage NF 66; Fig. 4O), the fore− and hind limbs are almost completely ossified except for the carpal and tarsal elements; the urostyle is present. Dermal ossifications of the skull are still weak, based on the fact that dermal bones of the skull are not preserved in the specimen SAM−PK−K9698. The iliac shafts are not entirely ossified, and also the sacral diapophyses, if present, are not yet ossified.

119

Adults (Fig. 4P, Q) may reach about 40 mm snout−vent length and are characteristic by ossified carpal and tarsal ele− ments and sacral diapophyses, the sacral vertebra co−ossified with the urostyle, and a completely ossified braincase (Fig. 4U). The braincase could be co−ossified with the frontoparietal (which is smooth on its dorsal surface, narrower in the middle of its length, and does not extend over the dorsal surface of the otic capsules; see SAM−PK−K9549a; SAM−PK−K9550a; SAM−PK−K 9593; SAM−PK−K9651; SAM−PK−K9668a; SAM−PK−K9733) and with the parasphenoid. On the other hand, the astragalus and calcaneum are not coalesced even in fully ossified adults (e.g., SAM−PK−K9513; SAM−PK−K9568, SAM−PK−K9653a, SAM−PK−K9666, SAM−PK−K9717a). The parasphenoid reaches the level of the symphysis of the lower jaw (SAM−PK−K 9549a; SAM−PK−K9658), but does not extend beyond the anterior margin of the frontoparietal (SAM−PK−K9669). The nasals are drop−shaped and moder− ately asymmetrical (SAM−PK−K9550a). The symphysis con− sists of the mentomandibulars (e.g., SAM−PK−9521; SAM− PK−9605; SAM−PK−K9653a). The jaw joint is located at the plane of the anterior portion of the otic capsules (SAM−PK− K9550a; SAM−PK−K9605; see also Fig. 4Q). The ribs are ar− ticulated with, but not co−ossified with, the transverse pro− cesses (SAM−PK−K9717b). The sacral diapophyses are mod− erately extended posteriorly (e.g., SAM−PK−K9625; SAM− PK−K9633; see also Fig. 4T). The vertebral centra are dis− tinctly opisthocoelous (e.g., SAM−PK−K9550a; SAM−PK− K9552; SAM−PK−K9637; SAM−PK−K9653a, SAM−PK− K9670a; SAM−PK−K9700; SAM−PK−K9707; see also Fig. 4Q). The vertebral column consists of eight presacrals, and vertebra 9 is the sacral (e.g., SAM−PK−K9576; SAM−PK− K9670; see also Fig. 4Q). The iliac shafts are straight, long, and extend nearly to the most posterior (3rd) rib (usually to vertebra 5) (e.g., SAM−PK−K9602; SAM−PK−K9610a). Spec− imen SAM−PK−K9687 provides valuable information on at least some carpal elements. The proximal row consists only of the ulnare and radiale; the radiale is distally adjoined by a comparatively small praeaxiale centrale, to which carpale distale 1 and 2 articulate. The praepollex is not preserved (ter− minology after Jarošová 1973). There is one individual (SEM−PK−K9559a, b), preserved in dorsal and ventral aspects (Fig. 4R, S), apparently in an ad− vanced metamorphic stage (judging by presence of unidenti− fiable ossification in the pelvic region), which differs from other specimens in having comparatively long, narrow and flat frontoparietal that lacks the typical median depression and by rather small otic capsules. Proportions of the skull suggest another pipid taxon at Stompoor. Similar proportions of the frontoparietal to the otic capsules were found in SAM−PK−K9669, which is a subadult (i.e., iliac shafts com− paratively short, sacral diapophyses not ossified) but in which the frontoparietal is comparatively narrow and long, and the parasphenoid, although well preserved, does not ex− tend beyond its anterior margin. Another tadpole (USP 493, Fig. 4V) of uncertain age, is from crater lake deposits of Zaire, and is mentioned here to http://app.pan.pl/acta51/app51−111.pdf

120

complete the list of otherwise scarce Mesozoic anuran larvae. It was found together with several adult frogs by employees of Zaire mining companies. The mining companies were reluc− tant to allow localities to be published, as it might help their competitors. This is why we only guess that these frogs, in− cluding the tadpole, are also of the Cretaceous age and related to the break−up of Gondwana, as the other pipids are thought to be. The material of adults (see also Van Dijk 1995: fig. 2) was lent to R. Estes in 1980s but now it is considered lost. Based on the number of vertebrae, presence of three pairs of ribs, and well ossified otic capsules, we estimate its devel− opmental stage as NF 60 (compare ossification of the otic capsules in Shomronella, Fig. 3C). It is preserved as imprints of the skeleton in a matrix, with some fragments of bones preserved in situ. Judging from the preserved part of vertebra 7 that includes zygapophyses, the specimen is exposed ven− trally. Ten ossified vertebrae can be seen, the most posterior (i.e., first caudal) being only an early rudiment. The otic cap− sules are well ossified and one can recognize the interior of the posterior semicircular canal and the prootic foramen. The most striking feature is the presence of three pairs of thin straight ribs, preserved as imprints in the matrix. Such “fish− like“ ribs are quite unusual for pipids.

Discussion There is a considerable variation in the appearance of ana− tomical features in the course of larval and metamorphic de− velopment of the Anura (see discussion in Roček 2003b). Variation is also caused by the preservation condition. For instance, non−skeletal structures are preserved in early larvae of Shomronella, which allow the reconstruction of their ex− ternal appearance and thus principal criteria for definition of developmental stages like those in contemporary Xenopus. However, because the degree of development of skeletal fea− tures may vary in individuals with the same external mor− phology, and because external morphology defines develop− mental stages, staging of fossil tadpoles preserved exclu− sively as ossified skeletons, may not be accurate. Moreover, ossification sequence inferred from histological examina− tions may considerably precede that inferred from cleared and stained specimens. Taken all this into account, develop− mental comparisons of fossil taxa among themselves and comparisons with contemporary relatives may not be quite precise. Nevertheless, general information on morphology and sequence of ossification may be inferred. Main developmental events are summarized in Table 2. They were selected mainly with regard to whether they may be observed in a majority of compared taxa. Obviously, in those taxa where only a few tadpoles (and thus few develop− mental stages) were available, the basis for comparisons is rather limited. Xenopus was used to represent contemporary pipids not only because its development is well known and is the basis for staging development in other anuran taxa, but mainly be−


cause of its free−living larvae. In all Cretaceous pipids, as well as in Paleogene/Neogene Palaeobatrachus, larvae were also free−living. “Direct development“ in Pipa is no doubt a secondary developmental strategy aquired comparatively recently. Free−living larvae were undoubtedly present in Palaeobatrachus, whose tadpoles could attain gigantic di− mensions (as evidenced by several specimens from the Up− per Vindobonian, Miocene, of Randecker Maar, Germany; Roček and Rage 2000). Tadpoles of all taxa included in this study, which possessed long tails even in late metamorphic stages, were undoubtedly free−living larvae as well. Sequence of ossification.—The earliest centres in contem− porary Xenopus are in the frontals, parasphenoid, and the posteromedial part of the otic capsules (exoccipitals). In our cleared−and−stained whole mounts the earliest traces of calci− fication can be observed in stages NF 54–57 (e.g., Fig. 3O), which approximately corresponds to data by Trueb and Han− ken (1992). Nieuwkoop and Faber (1967) mentioned the ear− liest ossification at the stage NF 51. Later appearance of the frontoparietal was recorded by Sedra and Michael (1957) who histologically examined wild caught specimens. Hence, their results are comparable to those by Nieuwkoop and Faber (1967) and confirm the view expressed by Estes et al. (1978) that there is greater variability in the timing of ossifi− cation in wild, than laboratory bred, populations. Disregard− ing these circumstances, one may conclude that in most cases the earliest dermal ossifications (represented by the frontals and parasphenoid) precede the endochondral ossifications (exoccipitals) and these, in turn, precede the ossification of the anterior neural arches. We cannot say much about Thoraciliacus and the South African pipid because their earliest larvae in the samples (Figs. 1A, 4A) obviously were more advanced than the earli− est stage of Shomronella (Fig. 2A1, A2). The earliest stage of Shomronella, however, displays a well defined parasphe− noid (except for its posterior part), which suggests its high degree of ossification, but there were no signs of frontals or exoccipitals. Judging by the shape of the body with a short tail and no legs, plus no traces of cartilages, one can con− clude that this stage is younger than the earliest stage with recorded ossification in contemporary Xenopus (see above). We roughly estimate the earliest Shomronella tadpoles to be in stages NF 47–50 because during this period of develop− ment in Xenopus the body shape in dorsal aspect remains generally the same. We conclude from this and other early larvae of Shomronella that dermal ossification began much earlier in development in this Early Cretaceous pipid than in contemporary Xenopus. Comparison of this earliest pre− served larva with two others, which are more developed (i.e., complete parasphenoid and ossified anterior neural arches), revealed that in Shomronella the frontoparietal be− gan to ossify later than the parasphenoid and probably si− multaneously with the anterior part of the vertebral column. Among the last ossification events in the sequence, which may be recognized on the fossil material are those associated


121

Parasphenoid well ossified

Stage 66

Stages 63–64

Stage 63

Stage 61

Stage 60

Stage 56

Stage 55

Stages 51–54

Developmental event

Stages 47–50

Table 1. Sequence of the main developmental events, which can be recognized in Shomronella jordanica.

×

Early frontoparietals

×

Frontoparietals reach level of anterior tip of parasphenoid

×

Frontoparietals fuse posteriorly

×

Frontoparietals fuse completely

×

Angulosplenials appear

×

Nasals appear

×

Beginnings of otic capsules

×

Posteromedial part of otic capsules well ossified

×

Otic capsules complete

×

Sphenethmoid of 2 ossifications Up to 8 pairs of neural arches

× ×

Up to 9 pairs of neural arches

×

Up to 10 pairs of neural arches

×

Anterior neural arches fuse

×

2 anterior vertebrae fuse

×

Formation of anterior centra

×

Hypochord present 3 pairs of ribs Femur, tibia and fibula appear

× × ×

Ilia appear

×

Ilia in contact with one another

×

Ilia in ultimate size and position

×

Femur in contact with ilium Suprascapula present Supposed coalescence of tibia and fibula

with the cranial and postcranial parts of the endoskeleton (e.g., completion of the ossification of the otic capsules or of the ilia). Most of these ossifications are completed well be− fore the end of metamorphosis (stage NF 63; see Table 2). Only the development of the urostyle is markedly delayed (a separate hypochord may be recognized in stage NF 66, whereas in Xenopus the cartilaginous neural arches within the urostyle may be recognized at the end of metamorphosis; Roček 2003b: fig. 3I). Fusion of the anterior two presacrals may vary. Whereas in Shomronella they obviously fused well before the end of metamorphosis (stage NF 60), they may still be separate in metamorphosed juvenile Thora− ciliacus (Trueb 1999). It should be noted that fusion of the anterior presacrals is not correlated with fusion of the sacral either with the posterior presacral or with the most anterior caudal. This is evidenced by Shomronella and Thoraciliacus in which anterior two (1st and 2nd) presacrals fuse with one another; in contrast, their sacrum does not fuse with neigh− boring vertebrae.

× × ×

The ossification sequence is remarkably stable in the anu− rans (with exception of dermal ossification of the jaws in lar− val pipids and Eleutherodactylus Duméril and Bibron, 1841 in which the jaw apparatus is deviated from a common scheme) for both fossil and recent taxa. Among the earliest bones are always the parasphenoid, pars frontalis of the frontoparietal, exoccipital and prooticum, whereas the pelvis and distal elements of the extremities (endochondral skeletal parts in general) are among the last—Pelobates cf. decheni Troschel, 1861 (see Maus and Wuttke 2004), Ascaphus truei Stejneger, 1899, Alytes obstetricans (Laurenti, 1768), Hyme− nochirus boettgeri (Tornier, 1896), Pipa myersi Trueb, 1984 (see Yeh 2002; Roček 2003a; and references therein). Timing of developmental appearance of diagnostic char− acters in various pipids.—The Cretaceous, Paleogene/Neo− gene, and contemporary pipids share a certain number of characters common to all of them, characters by which they can be distinguished from each other, and characters that dis− http://app.pan.pl/acta51/app51−111.pdf

122


? ? ? 59–60 63 63–64 ? ? 59 59–60

tinguish them from the palaeobatrachids. These characters appear at various developmental levels and may thus indicate their phylogenetic history. One of the features that are common to pipids and palaeo− batrachids is the relation between the parasphenoid and frontoparietal. In Thoraciliacus (Fig. 1C; see also Nevo 1968; Trueb 1999), Shomronella (Estes et al. 1978: fig. 2), South African pipid, Palaeobatrachus (Roček 2003b: fig. 3A, B), and Xenopus (Fig. 4M) it extends anteriorly beyond the level of the anterior margin of the frontoparietals. Ante− rior extent of the parasphenoid, together with the shape of its posterior part, which is clearly distinctive not only in adults (see Nevo 1968: fig. 10) but also in tadpoles (Figs. 1D1, 3A2), are characters arising in early development. To a certain ex−

59 59–60 60

< 59–60 60 60 63 < 63

? 51 ? ? 59 ? ? ? 51 ? 59 ? ? 51 62 ? < 59 ? ? ? ? 57–58 < 62 62 ? ? ? ?

58 58 58 59 60 64 64 60 59 59–60 60 ? 58 59 60 58 0 65 ? 60 < 58 59 59 60 63 63 59 64

Xenopus

< 59 ? ? 59 60–61 < ? ? 60–61 59 ? 60–61 ? ? < 59 60–61 59 0 66 < ? ? < 59 60–61 60–61 ? ? 63 ? 63

Palaeobatrachus

47–50 51–54 55 60 63 60 60 ? 51–54 56 63 63 51–54 56 60 60 60 0 63 61 55 60 61 63 66 63 60 63–64

Llankibatrachus

South African pipid

Parasphenoid well ossified Early frontoparietals Frontoparietals reach level of anterior tip of parasphenoid Frontoparietals fuse posteriorly Frontoparietals fuse completely Angulosplenials appear Nasals appear Clavicles present Early ossification of otic capsules Posteromedial part of otic capsules well ossified Otic capsules complete Sphenethmoid of 2 ossifications Up to 8 pairs of neural arches Up to 9 pairs of neural arches Up to 10 pairs of neural arches Anterior neural arches fuse 2 anterior vertebrae fuse with one another Sacral and praesacral vertebrae fuse Formation of anterior centra Hypochord present 3 pairs of ribs Femur appears Ilia appear Ilia in contact with one another Ilia in ultimate size and position Femur in contact with ilium Suprascapula present Supposed coalescence of tibia and fibula

Thoraciliacus

Developmental event

Shomronella

Table 2. Comparison of main developmental events in Mesozoic and Paleogene/Neogene pipids with free−swimming larvae, Palaeobatrachidae, and contemporary Xenopus (stages based on external characters according to Nieuwkoop and Faber 1967). ? – character cannot be observed on available material; 0 – character absent in the taxon. Data on Palaeobatrachus from Roček (2003b), data on Xenopus are based on material deposited in De− partment of Zoology, Charles University, Prague.

59 57–59 59 59 60 60 62 59–60 62 63 63 58 59 60 60 0 0 < 60 66 62 60 60 63 66 62–63 63 60–61

tent, this holds true also for the ultimate shape of the fronto− parietal. Presence of free ribs is another common feature of all pipids and palaeobatrachids: 3–4 pairs on vertebra 2–4/5 in pipids, 5 pairs on the anterior 5 vertebrae in Palaeobatrachus (Roček 2003b: figs. 1O, 2J). They develop, like all meso− dermal derivatives and thus in accordance with the somitic axial skeleton, in a rostrocaudal direction. In comparison, three pairs of ribs are present in Shomronella at stage NF 55, and in other Cretaceous and Paleogene/Neogene pipids as well as in Palaeobatrachus appear somewhat later: i.e., stages NF 59–60, stage NF 58. In contemporary Xenopus this is shifted to stage NF 62. In Palaeobatrachus a complete set of five pairs of ribs is recognizable in stages NF 59–60.


Difference in the number of ribs is difficult to explain in a phylogenetic sense. Larger numbers of ribs may be reason− ably considered a primitive state. In this respect, Palaeo− batrachus should be more primitive than all pipids. How− ever, there are undoubtedly five pairs of ribs in one Shom− ronella tadpole (HUJZ 12038; see Estes et al. 1978: fig. 7A) and regularly four pairs of ribs in Early Cretaceous adult Thoraciliacus and Cordicephalus (Nevo 1968; Trueb 1999). These suggest a decrease in number of ribs during the early evolution of pipoids and, therefore, larger numbers of ribs should be expected in ancestral pipoids earlier than Shom− ronella. Because palaeobatrachids retain five pairs of ribs, they supposedly had to split from the common pipoid stem before Shomronella (this would also correspond to molecu− lar data indicating earlier, probably Laurasian origin of the common stock; Feller and Hedges 1998). This is, however, not confirmed by the fossil record in which the earliest palaeobatrachids are recorded only from the Late Cretaceous (Campanian and Maastrichtian) of France and Spain, respec− tively (Buffetaut et al. 1996; Astibia et al. 1990). The gap be− tween primitive hypothetical Early Cretaceous pipoids and the earliest known palaeobatrachids is about 45 Ma. One thing should be emphasized in considering ribs and their significance for phylogenetic reconstructions. The only other Mesozoic anurans that possess free ribs are disco− glossoids. They also have three pairs of free ribs on vertebrae 2–4, but the second pair each bear a distinct uncinate process. This characteristic feature was maintained even in the Paleo− gene/Neogene discoglossid Latonia (Roček 1994) and in contemporary Discoglossus. It was also present in the earli− est anuran Prosalirus from the Jurassic (Pliensbachian) of Arizona (Jenkins and Shubin 1998) and in Vieraella and Notobatrachus from the Early Jurassic (Liassic and the Callovian–Oxfordian, respectively) of South America (Báez and Basso 1996). Presence of the uncinate process on the ribs associated with vertebra 3 is therefore a primitive feature that is maintained only in the discoglossoid lineage (present also in contemporary Ascaphus and Leiopelma) but not in the pipoid lineage – with the exception of Shomronella (Fig. 3E). This may be taken as an evidence of phylogenetic relations of early pipoids to the discoglossoid lineage, and in general, a primitive feature retained in the most primitive anurans. Comparison of developments in Cretaceous pipids and in Palaeobatrachus.—Comparisons of the beginnings of dermal ossification in Shomronella (and, curiously, also in Paleogene Llankibatrachus) revealed that they distinctly preceded those in Palaeobatrachus. This can be shown by the onset of ossifi− cation of most cranial elements (Table 2). In contrast, the peri− ods during which single cranial elements (both dermal and chondral) developed were considerably longer in Shomronella than in Palaeobatrachus. For instance, the frontoparietals of Shomronella appeared in stages NF 51–54 and completed their fusion in stage NF 63, whereas in Palaeobatrachus they appeared as late as stage NF 58, but were completely fused as early as stage NF 60. Similarly, the otic capsules of Shomro−

123

nella begin to ossify in stages NF 51–54 and are completely ossified in stage NF 63, whereas in Palaeobatrachus they be− gin to ossify in stage NF 59 and are complete in stage NF 60. In Thoraciliacus, the situation is obscured by the lack of rele− vant specimens. It may be of some interest that the otic cap− sules seem to begin ossification rather late (i.e., stage NF 59; Fig. 1A) and are complete as early as stages NF 60–61. It is difficult to decide whether this was caused by incomplete preservation (stage NF 59 in the earliest available stage for Thoraciliacus) or if it reflects the real situation. The latter seems to be supported by Stompoor pipid development. Regarding the postcranial skeleton, Shomronella has eight pairs of neural arches by stages NF 51–54 and 10 pairs in stage NF 60, whereas Palaeobatrachus has eight pairs of neural arches much later, at stage NF 58, and 10 pairs at stage NF 60. This means that also in the axial skeleton the develop− ment of Palaeobatrachus was abbreviated. Remarkably, the hind limbs develop later in Shomronella than in Palaeobatrachus, which is the reverse of the situation to that in the cranium. However, like the skull and axial skel− eton, also the hind limbs have longer development in Sho− mronella than in Palaeobatrachus. In Shomronella, the fe− mur appears in stage NF 60 and the ilia are in their ultimate size and position only in stage NF 66, whereas in Palaeo− batrachus the femur appears in stage NF 59, but iliac devel− opment ends as early as stage NF 63. Shape of the vertebral centra is considered an important feature distinguishing pipids and palaeobatrachids from each other (opisthocoelous in the former, procoelous in palaeo− batrachids). Procoelous shape of the vertebrae in Palaeo− batrachus is established as early as stage NF 59 (Roček 2003b: fig. 4H). In Shomronella, the vertebral centra begin to ossify above the notochord (epichordal pattern according to Estes et al. 1978) so that the centra are incomplete ventrally in stage NF 63, indicating the presence of the notochordal canal. In the horizontal section, the centra at this stage are amphicoe− lous (see also Estes et al. 1978: fig. 6C). Soon, however, the notochordal canal is filled with calcified matrix, first in the an− terior vertebrae and then added more caudally, and inter− vertebral spaces are filled with ovoid elements made of the same material. These ovoid elements soon fuse with the calci− fied matrix filling the notochordal canal of the adjacent poste− rior vertebrae. During later development (stage NF 63), these calcifications ossified and coalesced with earlier vertebral ru− diments, giving rise to the opisthocoelous centrum. This pro− cess is shown in fig. 6C of Estes et al. (1978). It is also obvious that the opisthocoelous type of the Early Cretaceous Shomro− nella appears later in development than the procoelous type in the late Oligocene Palaeobatrachus, but in light of other heterochronic shifts this is not so important. It is more impor− tant that the procoelous condition of early palaeobatrachids could hypothetically arise in a similar way, when the ovoid calcification filling the intervertebral space could join and fuse with the vertebral segment adjacent anteriorly. Comparison of Palaeobatrachus with Xenopus revealed that, whereas the beginnings of the ossification are about at http://app.pan.pl/acta51/app51−111.pdf

124

the same stage (NF 58–59) and both the parasphenoid and frontoparietals appear at about the same time, the dermal bones that develop later (e.g., angulosplenials, nasals) appear much later in Palaeobatrachus than in Xenopus. General features of evolution of amphibian ontogenies.— There is some information available about the development of the primitive amphibians, namely of the Branchiosauridae (Temnospondyli) from the Lower Permian of Germany (Boy 1974; Schoch 1998), that can be compared with development of Cretaceous as well as contemporary pipids. Data on deve− lopment of primitive recent urodeles [Ranodon sibiricus Kessler, 1866, Salamandrella keyserlingi Dybowski, 1870, Ambystoma mexicanum (Shaw, 1789)] may be used for such comparisons as well (Lebedkina 2004). The main constraints in these comparisons is that one can use only relative se− quence of developmental events, mainly onset of ossifica− tion, and not of corresponding developmental stages. Rarely can fossil amphibians be definitely staged. This restricts comparisons to ossification sequence only. Despite this constraint, the parasphenoid is always the earliest, or among the earliest, cranial ossifications in Paleo− zoic branchiosaurids (see Schoch 1998: fig. 8 for Apateon pedestris Meyer, 1844; Boy 1974: fig. 1 and table 1 for Branchiosaurus sp.). However, onset of ossification of the parasphenoid is accompanied by vomers, praemaxillaries and pterygoids. The frontals, which appear always before the parietals in the anurans (Čihák et al. 2003; Lebedkina 2004), appear much later in the branchiosaurids; i.e., only after the pterygoid, maxilla, and palatine, which appear one after an− other. In the branchiosaurids, the parietals and squamosals appear simultaneously with the frontals, and are then fol− lowed by the quadratojugals shortly before metamorphosis, and the nasals at the beginning of metamorphosis. The septo− maxillary appears only at the end of metamorphosis. It should be noted that in Branchiosaurus the neural arches ap− pear rather early, shortly after the parasphenoid (but much earlier than the frontals), whereas the ilia appear only in late metamorphosis (Boy 1974). One can suppose that this devel− opmental scheme might correspond to that in ancestors of the Anura and of the Caudata. In the Caudata, the sequence of ossification is slightly dif− ferent. In the earliest stage, when the mouth forms in Rano− don, the dentary and coronoid appear as in Branchiosaurus, but they are followed by the parasphenoid, three pairs of frontals, parietals, vomers and palatines only in the subse− quent stage (stage III of Lebedkina; but cf. Schoch 1998: fig. 8). The preoperculars, squamosals and angulars appear only with the transition to active feeding, and the exoccipitals ap− pear still later. In Salamandrella, the earliest bones to appear are again the dentary, coronoid, vomer and palatine, whereas the parasphenoid appears much later (with active feeding), followed still later by two posterior pairs of frontals and by the parietal. Only when these bones appear does the ex− occipital begin to ossify. In Ambystoma, the onset of ossifica− tion sequence is the same as in Salamandrella, soon followed


by the parasphenoid and the anterior pair of frontals (besides other bones), with the two posterior pairs of frontals appear− ing later, and parietals still later. It is obvious that the begin− nings of the ossification sequence (e.g., vomer, dentary, coronoid) is basically the same as in the Branchiosauridae, and only few elements (e.g., maxilla) are shifted to later stages. In the anurans (including Early Cretaceous pipids) the skeleton is simplified and the whole ossification sequence is shifted to metamorphosis. The definition of metamorpho− sis in the temnospondyls and caudates, principally based on disappearance of gill−rakers that reflects a transition from gill breathing to lung breathing, differs from that in the anu− rans, because of external criteria available. In the anurans, the larval branchial skeleton is converted into a filter−feed− ing device, nevertheless, the beginning of metamorphosis in all groups can be defined with appearance of limbs in a larva (tadpole). The criterion of its end differs—in the temnospondyls and caudates it is a complete disappearance of gill rakers and in the anurans it is the complete disappear− ance of the tail. However, it is well known that in the anu− rans the transformation of the branchial skeleton into the hyoid, as well as its ossification (both dermal and endo− chondral) continues postmetamorphically (Roček 1981, 2003a). Despite these differently defining events associated with metamorphosis, it is obvious that the skeletal develop− ment in frogs is shifted to metamorphosis. This would be even more apparent if appearance of the limbs would be taken as the common criterion of the beginning of metamor− phosis in all groups. It may be inferred from comparing temnospondyls with anurans (see Fig. 5) that in the former group there was a markedly longer ossification period extending from initial mouth formation until the adult. In the anurans, in contrast, the ossification period was shorter and occurring in later de− velopmental stages. Seemingly, this could have been caused by abbreviation of the ossification period and its shift to later developmental stages. In fact, this heterochrony can also be caused by a shift of metamorphosis to earlier developmental stages, which is evidenced, among other features, by the presence of juvenile teeth of temnospondyls in the adult anu− rans (Smirnov and Vasil’eva 1995), and which would better correspond to the idea of paedomorphosis (defined as accel− eration of development accompanied with abbreviated somatogenesis). In pipids, whose developmental sequence is recorded from the Early Cretaceous (i.e., more than 120 Ma), the situation is similar—onset of ossification is seemingly postponed to later developmental stages (if Xenopus is com− pared with Shomronella) while its concluding phases occur in the corresponding stages. This means that the whole ossifi− cation process is abbreviated into a shorter period. This could, however, also mean that metamorphosis was shifted to earlier developmental stages in contemporary Xenopus (see also Wassersug and Hoff 1982). Delayed dermal ossification of the jaws in the Anura (Yeh 2002), compared with the Paleozoic Temnospondyli


and recent Caudata (Fig. 5), is no doubt caused by adaptation to herbivory during their larval period. The herbivory of the anuran larvae is anatomically associated with the supra− rostral and infrarostral cartilages (i.e., larval jaws) that are supported by enlarged horizontal palatoquadrates and by Meckel’s cartilage (see Roček 2003a for review). Only dur− ing metamorphosis do these larval jaws become incorporated into the nasal region (suprarostrals) and the adult lower jaw (infrarostral, which ossifies into the mentomandibular), and the principal elements of the jaw apparatus in the adult anu− rans are the palatoquadrate and Meckel’s cartilage, respec− tively, accompanied with their adjacent dermal ossifications. In pipids the larval jaw apparatus is not developed (due to ab− sence of larval stages adapted to herbivory that occur in non−pipids) and this is the reason why their dermal ossifica− tion of the jaws seems to occur earlier than in non−pipids, and similar to the caudates. It would be therefore interesting to know more about the larval development and metamorphosis in the Mesozoic non− pipid anurans (e.g., early discoglossoids such as Eodisco− glossus), namely whether their larval cranial structure devi− ated from the general scheme of larval temnospondyls as well as of pipids because of adaptation to herbivory (these devia− tions principally include anterior rotation of the palatoqua− drate and presence of labial cartilages that precede jaws of adults; see Roček 2003a for a review). There is a representa− tive developmental series of Eopelobates from the Upper Oligocene of the Czech Republic (Špinar 1972), and about 70 fossil tadpoles of Pelobates cf. decheni Traschel, 1861 from the Upper Oligocene of Germany (Maus and Wuttke 2004), which provide fairly good knowledge on the ossification se− quence. These specimens come from the Paleogene, so that they are basically identical with contemporary pelobatids. However, they do not involve larval cartilaginous endocranial structures that are of key importance in this respect. Oligocene pelobatids are thus too advanced to provide insight into the de− velopment of the early non−pipoid anurans. Summary.—The earliest ossified element in the Early Cre− taceous Shomronella is the parasphenoid which may be well ossified as early as in stages NF 47–50. In contrast, in the contemporary Xenopus and in the Late Oligocene Palaeo− batrachus, the first ossification can be recognized only in stages NF 59 and 58, respectively. This means that the onset of ossification was shifted into later developmental stages in the course of pipid evolution. · The differences between the Early Cretaceous taxa and contemporary Xenopus are more pronounced in those os− sification events that occur comparatively early (e.g., ap− pearance of the paraspenoid), whereas in those which oc− cur in later stages (e.g., angulosplenials and nasals) the dif− ference is less pronounced. · Among the last ossification events, the coalescence of the tibia and fibula, occurs in stages NF 63–64 in Shomronella and in 63 in Thoraciliacus, but occurs in stages NF 60–61 in Xenopus. Ossification thus extended over a longer period in

·

·

·

· ·

·

·

125

the development of early pipids than in the Paleogene palaeobatrachids and contemporary Xenopus. As a consequence of faster ossification of the skeleton, the ossification of the elements, which was sequential (e.g., parasphenoid before the frontoparietals, frontoparietals before the exooccipitals) became less pronounced and some elements ossify at the same stage. The developmental sequence for the ossification of the parasphenoid, frontoparietals, and exoccipitals, which oc− curs in that order, in the Early Cretaceous Shomronella, tend to be reversed in some Xenopus individuals (i.e., frontoparietals before parasphenoid). These heterochronic shifts in developmental timing con− cern mainly the skull and anterior vertebrae. The posterior vertebrae, hypochord, and appendages do not exhibit these heterochronic changes. This shift in ossification sequence is not obvious if Creta− ceous and Paleogene/Neogene pipids are compared to each other. The ribs associated with the vertebra 3 in Shomronella bear a distinct uncinate process, similar to this in the Jurassic anurans Prosalirus, Vieraella and Notobatrachus, as well as contemporary discoglossoids Ascaphus and Leiopelma. Their presence in Shomronella might indicate phylogenetic relations of early pipoids. Development in Shomronella elucidates the mechanism for the origin of the opisthocoelous vertebral centra. The same mechanism, if directionally reversed, might explain the origin of the procoelous centra in Palaeobatrachus. Comparisons of the development in temnospondyl amphib− ians with that of pipoid anurans indicates that in the anurans the whole ossification sequence is shifted from pre−meta− morphic stages to metamorphosis. This also could be inter− preted in the context of the hypothesis of the paedomorphic origin of the anurans; i.e., seemingly the metamorphosis of frogs, including their entire ossification sequence, was shifted to the earlier developmental stages, due to overall abbreviation of their somatogenesis.

Acknowledgements We want to express our gratitude to Rivka Rabinovich and Yehudah Werner (Department of Evolution, Systematics and Ecology, the Hebrew University of Jerusalem), for their hospitality and assistance they kindly offered to ZR. Thanks are due also to Roger Smith (South African Mu− seum, Cape Town) and Boris Ekrt (National Museum, Prague) for allow− ing us to investigate specimens under their care. We are indebted to Ron Altig (Mississippi State University, Mississippi State city) who kindly provided helpful comments on an earlier draft of the manuscript, and to Gregory R. Handrigan and Richard Wassersug (Dalhousie University, Halifax), and Jean−Claude Rage (Muséum d’Histoire naturelle, Paris), for their constructive suggestions. This research was made possible by the grant No A3013206 from the Grant Agency of the Academy of Sci− ences of the Czech Republic and from the National Research Foundation of South Africa to ZR, and AVOZ30130516 of the Geological Institute, Academy of Sciences of the Czech Republic. http://app.pan.pl/acta51/app51−111.pdf

126

References Astibia, H., Buffetaut, E., Buscalioni, A.D., Cappetta, H., Corral, C., Estes, R., Garcia−Garmilla, F., Jaeger, J., Jimenez−Fuentes, E., Le Leuff, J., Mazin, J.M., Orue−Etxebarria, X., Pereda−Suberbiola, J., Powell, J.E., Rage, J.C., Rodriguez−Lazaro, J., Sanz, J.L., and Tong, H.1990. The fossil vertebrates from Laño (Basque Country, Spain); new evidence on the composition and affinities of the Late Cretaceous continental faunas of Europe. Terra Nova 2: 460–466. Báez, A.M. and Basso, N.G. 1996. The earliest known frogs of the Jurassic of South America: Review and cladistic appraisal of their relationships. Münchner geowissenschaftliche Abhandlungen (A) 30: 131–158. Báez, A.M. and Púgener, L.A. 2003. Ontogeny of a new Palaeogene pipid frog from southern South America and xenopodinomorph evolution. Zoological Journal of the Linnean Society 139: 439–476. Boy, J.A. 1974. Die Larven der rhachitomen Amphibien (Amphibia: Temno− spondyli; Karbon – Trias). Paläontologische Zeitschrift 48: 236–268. Buffetaut, E., Costa, G., Le Loeuff, J., Martin, M., Rage, J.−C., Valentin, X., and Tong, H. 1996. An Early Campanian vertebrate fauna from the Villeveyrac Basin (Hérault, Southern France). Neues Jahrbuch für Geologie und Paläontologie, Mitteilungen 1996: 1–16. Chipman, A.D. and Tchernov, E. 2002. Ancient ontogenies: larval develop− ment of the Lower Cretaceous anuran Shomronella jordanica (Amphibia: Pipoidea). Evolution and Development 4: 86–95. Čihák, R., Královec, K., and Roček, Z. 2003. Developmental origin of the frontoparietal bone in Bombina variegata (Anura: Discoglossidae). Journal of Morphology 255: 122–129. Daudin, F.M. 1802. Histoire naturelle des rainettes, des grenouilles et des crapauds. 71 pp. Levrault, Paris. Duméril, A.M.C. and Bibron, G. 1841. Erpétologie Général ou Histoire Naturelle complète des Reptiles. Vol. 8. 784 pp. Roret, Paris. Dybowski, B. 1870. Beitrag zur Kenntniss der Wassermolche Sibiriens. Verhandlungen der zoologisch−botanischen Gesellschaft in Wien 20: 237–242. Estes, R., Špinar, Z.V., and Nevo, E. 1978. Early Cretaceous pipid tadpoles from Israel (Amphibia: Anura). Herpetologica 34: 374–393. Feller, A.E. and Hedges, S.B. 1998. Molecular evidence for the early history of living amphibians. Molecular Phylogenetics and Evolution 9: 509–516. Gvirtzman, G., T. Weissbrod, G. Baer, and Brenner, G. J. 1996. The age of the Aptian Stage and its magnetic events: new Ar−Ar ages and palaeomagnetic data from the Negev, Israel. Cretaceous Research 17: 293–310. Jarošová, J. 1973. The components of the carpus in Palaeobatrachus and their development in two related Recent species. Časopis Národního muzea, oddíl přírodovědný 142: 89–104. Jenkins, F.A. Jr. and Shubin, N.H. 1998. Prosalirus bitis and the anuran caudopelvic mechanism. Journal of Vertebrate Paleontology 18: 495– 510. Kessler, K.F. 1866. Beschreibung eines neuen Wassermolches aus West− Sibirien. Bulletin de la Société Impériale des Naturalistes de Moscou 39: 126–131. Lang, B. and Mimran, Y. 1985. An early Cretaceous volcanic sequence in central Israel and its significance to the absolute date of the base of the Cretaceous. Geology 93: 179–184. Laurenti, J.N. 1768. Specimen medicum, exhibens Synopsin Reptilium emendatum cum Experimentis circa Venena et Antidota Reptilium Austriacorum. 214 pp. J.T. de Trattnern, Wien. Lebedkina, N.S. 2004. Evolution of the amphibian skull. Advances in Am− phibian Research 9, 1–266. Pensoft, Sofia. Maus, M. and Wuttke, M. 2004. The ontogenetic development of Pelobates cf. decheni tadpoles from the Upper Oligocene of Enspel (Westerwald/Ger− many). Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen 232: 215–230. Meyer, H. von 1844. Briefliche Mittheilung an Prof. Bronn. Neues Jahrbuch für Mineralogie, Geognosie, Geologie und Petrefaktenkunde 1844: 336– 337.

ACTA PALAEONTOLOGICA POLONICA 51 (1), 2006 Nevo, E. 1956. Fossil frogs from a Lower Cretaceous bed in southern Israel (central Negev). Nature 178: 1191–1192. Nevo, E. 1968. Pipid frogs from the Early Cretaceous of Israel and pipid evolu− tion. Bulletin of the Museum of Comparative Zoology 136: 255–318. Nieuwkoop P.D. and Faber, J. 1967. Normal table of Xenopus laevis (Daudin). 238 pp. North−Holland Publishing Company, Amsterdam. Roček, Z. 1981. Cranial anatomy of frogs of the family Pelobatidae Stannius, 1856, with outlines of their phylogeny and systematics. Acta Universitatis Carolinae – Biologica 1980: 1–164. Roček, Z. 1994. Taxonomy and distribution of Tertiary Discoglossids (Anura) of the Genus Latonia v. Meyer, 1843. Geobios 27: 717–751. Roček, Z. 2003a. Larval development and evolutionary origin of the anuran skull. Chapter 7. In: H. Heatwole and M. Davies (eds.), Amphibian Biol− ogy. Vol. 5: Osteology, 1877–1995. Surrey Beatty & Sons, Chipping Norton, Australia. Roček, Z. 2003b. Larval development in Oligocene palaeobatrachid frogs. Acta Palaeontologica Polonica 48: 595–607. Roček, Z. and Rage, J.−C. 2000. Tertiary Anura of Europe, Africa, Asia, North America, and Australia. Chapter 15. In: H. Heatwole and R.L. Carroll (eds.), Amphibian Biology. Vol. 4: Palaeontology, 1332–1387. Surrey Beatty & Sons, Chipping Norton, Australia. Schoch, R.R. 1998. Homology of cranial ossifications in urodeles: Signifi− cance of developmental data for fossil basal tetrapods. Neues Jahrbuch für Geologie und Paläontologie, Mitteilungen 1998: 1–25. Sedra , S.N. and Michael, M.I. 1957. The development of the skull, visceral arches, larynx and visceral muscles of the South African Clawed Toad, Xenopus laevis (Daudin) during the process of metamorphosis (from Stage 55 to Stage 66). Verhandelingen der Koninklijke Nederlandse Akademie van Wetenschappen, afdelningen Natuurkunde 51: 1–80. Shaw, G.and Nodder, F.P. 1789. The Naturalist’s Miscellany, or Coloured Figures of Natural Objects; Drawn and Described from Nature.Vol. 9. (Unnumbered pages). London. Smirnov, S.V. and Vasil’eva, A.B. 1995. Anuran dentition: development and evolution. Russian Journal of Herpeology 2: 120–128. Smith, R., Ross, C., and Asher, R. 2002. Taphonomy and palaeoenvironments of Late Cretaceous anurans and dinosaurs from a crater lake deposit in South Africa. Abstracts of the 8th International Symposium Mesozoic Ter− restrial Ecosystems 2002, p. 39. Cape Town. Špinar, Z.V. 1972. Tertiary Frogs from Central Europe. 253 pp. Academia, Prague. Stejneger, L. 1899. Description of a new genus and species of Discoglossid toad from North America. Proceedings of the United States National Museum 21: 899–900. Tornier, G. 1896. Die Kriechthiere Deutsch−Ost−Afrikas. Beiträge zur Syste− matik und Deszendenzlehre. xiii + 164 pp. Geographische Verlagsbuch− handlung Dietrich Reimer, Berlin. Troschel, F.H. 1861. Uebersicht aller bisher aus der Braunkohle des Sieben− gebirges beschriebenen fossilen Tiere. Sitzungsberichte der niederrheini− schen Gesellschaft zu Bonn 1861: 55–56. Trueb, L. 1984. Description of a new species of Pipa (Anura: Pipidae) from Panama. Herpetologica 40: 225–234. Trueb, L. 1999. The early Cretaceous pipoid anuran, Thoraciliacus: redescrip− tion, revaluation, and taxonomic status. Herpetologica 55: 139–157. Trueb, L. and Hanken, J. 1992. Skeletal development in Xenopus laevis (Anura: Pipidae). Journal of Morphology 214: 1–41. Trueb, L., Ross, C.F., and Smith, R. 2005. A new pipoid anuran from the Late Cretaceous of South Africa. Journal of Vertebrate Paleontology 25: 533–547. Van Dijk, D.E. 1985. An addition to the fossil Anura of Southern Africa. South African Journal of Science 81: 207–208. Van Dijk, D.E. 1995. African fossil Lissamphibia. Palaeontologia Africana 32: 39–43. Wassersug, R.J. and Hoff, K. 1982. Developmental changes in the orienta− tion of the anuran jaw suspension: a preliminary exploration into the evolution of metamorphosis. Evolutionary Biology 15: 233–246. Yeh, J. 2002. The evolution of the development: two portraits of skull ossifi− cation in pipoid frogs. Evolution 56: 2484–2498.